An Efficient Algorithm for Influence Maximization under Linear Threshold Model

Influence maximization is to find a small set of most influential nodes in the social networks to maximize their aggregated influence in the network. The high complexity of the classical greedy algorithm cannot be well suited for the moderate or large scale networks. It is necessary to develop a more efficient algorithm, not sensitive to the scale of the social network. In this paper, we propose an approach for estimating the nodes' influence based on the network structure. By this way, we make the scope of influence reduced to the nodes with the maximal influence, while make the consuming time reduced consequently. Then, we design a more efficient greedy algorithm (called LNG algorithm) for the linear threshold model. Experimental results on large scale networks demonstrate that the time consuming is much less and the influence spread effect is better than the classical greedy algorithm.